Fig. 3

Screening of the optimal prognostic gene set for osteosarcoma and the receiver operating characteristic (ROC) curve of SVM classification. a: The identification of optimal prognostic gene set for osteosarcoma based on the recursive feature elimination algorithm. The horizontal axis shows the number of differentially expressed genes and the vertical axis represents the cross-validation accuracy. b: The ROC curve of SVM classification in training dataset. c: The ROC curve of SVM classification in validation dataset